Fuzzy backward reasoning device

• US 5,289,563 A
• Filed: 05/22/1991
• Issued: 02/22/1994
• Est. Priority Date: 03/08/1990
• Status: Expired due to Fees

• Current Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Original Assignee: Mitsubishi Denki Kabushiki Kaisha (Mitsubishi Electric Corporation)
• Inventors: Kirimoto, Tetsuo, Nomoto, Kohei, Ohhashi, Yoshimasa
• Primary Examiner(s): Macdonald, Allen R.
• Assistant Examiner(s): Downs, Robert W.
• Application Number:
  US07/704,262
• Time in Patent Office: 1,007 Days
• Field of Search: 395/61, 395/51, 395/900, 395/902, 395/50, 364/516, 382/14, 382/15
• US Class Current: 706/45
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G06N 7/023 Learning or tuning the parameters of a fuzzy system
• CPC Class Codes: Y10S 706/90 Fuzzy logic
• CPC Class Codes: Y10S 706/902 Application using ai with detail of the ai system

Abstract

A sequential type Fuzzy backward reasoning device is disclosed, which is capable of performing computation progressively every time a feature quantity is observed to update reasoning, and of performing the reasoning even if the order of observations is arbitrary, by providing means for performing the sequential reasoning instead of batch type reasoning means, and feedback means for feeding back a reasoned result.

Additionally, a target recognition device is disclosed, which is capable of computing the reliability of a recognized result on a target and outputting the same as a numerical value, by obtaining another recognition information even if there is not obtained any information concerning the target, using said sequential type Fuzzy backward reasoning device.

Claims

1. A target recognition device comprising a sensor for observing a target and for obtaining information concerning the target, a plurality of feature quantity extractors for receiving the information concerning the target outputted from said sensor and for extracting a feature quantity of the target, and a Fuzzy backward reasoning device for receiving the information of the feature quantity outputted from each feature quantity extractor and for reasoning the kind of the target by Fuzzy backward reasoning based upon a Fuzzy theory from kinds of targets known as candidates and knowledge information concerning those targets.

2. A target recognition device according to claim 1 characterized in that it includes a sequential type Fuzzy backward reasoning device as said Fuzzy backward reasoning device, and a selector for receiving outputs from the plurality of feature quantity extractors and for extracting an activated feature quantity among said outputs for an input to said sequential type Fuzzy backward reasoning device.

3. A target recognition device according to claim 1 or 2 characterized in that it includes a radar as the sensor, and a tracking filter for receiving an output from said radar and for estimating and outputting the accurate position and speed of the target.

4. A method of performing Fuzzy backward reasoning comprising the steps of:

one or more sensors sensing a sequence of observed feature quantities representative of a cause;

said one or more sensors supplying said sequence of observed feature quantities to a computer;

performing a sequence of Fuzzy backward reasoning steps in the computer during sequential reasoning periods in response to said sequence of observed feature quantities, each Fuzzy backward reasoning step being based on a feature quantity of said sequence of observed feature quantities, a causality relation between said feature quantity and causes of said feature quantity, and a previous reasoned result from a previous Fuzzy backward reasoning step, said causality relation being stored in said computer, each Fuzzy backward reasoning step outputting a reasoned result indicating a cause of the feature quantity;

combining a predetermined number of sets of reasoned results among said reasoned results outputted sequentially to form a union as one reasoned result; and

supplying the union to the next Fuzzy backward reasoning step as a previous reasoned result.

5. A method of performing Fuzzy backward reasoning comprising the steps of:

one or more sensors sensing a sequence of observed feature quantities representative of a cause;

said one or more sensors supplying said sequence of observed feature quantities to a computer;

performing a sequence of Fuzzy backward reasoning steps in the computer during sequential reasoning periods in response to said sequence of observed feature quantities, each Fuzzy backward reasoning step being based on a feature quantity of said sequence of observed feature quantities, a causality relation between said feature quantity and causes of said feature quantity, and a previous reasoned result from a previous Fuzzy backward reasoning step, said causality relation being stored in said computer, said feature quantity and said causality relation being received as interval Fuzzy quantities for said Fuzzy backward reasoning step, each Fuzzy backward reasoning step outputting a reasoned result indicating a cause of the feature quantity; and

supplying the reasoned result of each Fuzzy backward reasoning step to the next Fuzzy backward reasoning step as a previous reasoned result.